Issue #722: Document restrictions on types IH and EH in PSA

[jafingerhut]

No description provided.

[jafingerhut]

Intended to address this issue #722

[hesingh]

Why does section 7.2.7 of p4-16 working doc does not allow header_union as a member of struct?

[jafingerhut]

@hesingh If I go to the working draft of the P4_16 specification here: https://p4.org/p4-spec/docs/P4-16-working-spec.html#sec-type-nesting the first table I see has a row labeled "header_union", and in the column labeled "struct or tuple" the table entry says "allow". Do you see something different, or interpret that somehow as not being allowed?

[hesingh]

I see

struct | allowed2 | error | allowed

7.2.7. Type nesting rules

The table below lists all types that may appear as members of headers, header unions, structs, and tuples. Note that int means an infinite-precision integer, without a width specified.

  |   |   |  
bit | allowed | error | allowed
int | allowed | error | allowed
varbit | allowed | error | allowed
int | error | error | error
void | error | error | error
error | error | error | allowed
match_kind | error | error | error
bool | allowed | error | allowed
enum | allowed1 | error | allowed
header | error | allowed | allowed
header stack | error | error | allowed
header_union | error | error | allowed
struct | allowed2 | error | allowed
tuple | error | error | allowed
  |   |   |  

Rationale: int does not have precise storage requirements, un

7.2.7. Type nesting rules The table below lists all types that may appear as members of headers, header unions, structs, and tuples. Note that int means an infinite-precision integer, without a width specified

[jafingerhut]

This row in that table:

struct | allowed2 | error | allowed

means:

struct is allowed as a member in a header (the column heading of the second column), but see footnote 2.
it is an error if you attempt to use a struct as a member of a header_union
struct is allowed as a member in a struct or tuple

[apinski-cavium]

Note something is going on with the draft HTML generation though, see #1175 . The released spec 1.2.3 is correct.

[hesingh]

@jafingerhut Duh, got it, sorry. Your proposal looks fine.